Platform-Level Indicators of Application Performance

1. A method for monitoring performance, comprising: communicating application traffic comprising requests and responses of a single computer application that is carried in inbound and outbound packet traffic via at least one socket over a data network, the at least one socket excluding packet traffic of other computer applications; monitoring receipt times and exit times of data packets and respective packet headers to identify corresponding request packets and response packets of the single computer application among packets of the inbound and outbound packet traffic transiting the at least one socket; defining a performance indicator of the application traffic based on the receipt times and exit times of the corresponding request packets and response packets; measuring a first instance of the performance indicator during a first monitoring interval; measuring a second instance of the performance indicator during a second monitoring interval; making a determination that a change between the first instance and the second instance exceeds a predetermined threshold; and responsively to the determination, issuing an alert.

2. The method according to claim 1 , wherein the application runs on a virtual machine.

3. The method according to claim 1 , wherein the performance indicator comprises an average request arrival rate that is derived from the request packets.

4. The method according to claim 1 , wherein the performance indicator comprises an average queue time obtained by applying Little's law to a measurement of the request packets and response packets.

5. The method according to claim 1 , wherein the performance indicator comprises a mean latency obtained from differences in time between sending a request in the packet traffic and receiving a response.

6. The method according to claim 5 , wherein the performance indicator comprises a mean service time obtained as a difference between the mean latency and an average queue time obtained by applying Little's law to a measurement of the request packets and response packets.

7. The method that according to claim 1 , wherein the performance indicator comprises a frequency distribution of at least one of a mean service time and a mean queue time of the request packets and response packets.

8. The method according to claim 1 , wherein the steps of measuring a first instance and measuring a second instance are performed only when the application traffic exceeds a predetermined level.

9. The method according to claim 1 , further comprising the step of adjusting a computer resource responsively to the determination.

10. The method according to claim 9 , wherein adjusting the resource of the computer comprises automatically adjusting the resource of the computer.

11. The method according to claim 1 , further comprising the steps of: recording data of the request packets and response packets; and identifying packet trains of multipacket messages among the request packets and response packets of the single computer application, the packet trains comprising a plurality of entrained packets transmitted in succession, wherein recording data is performed for exactly one packet of respective packet trains and recording data is not performed for other packets thereof, and wherein measuring the first instance and measuring the second instance of the performance indicator are performed responsively to the step of recording data.

12. The method according to claim 11 , wherein the performance indicator is an average interarrival rate of the requests and responses.

13. A computer software product for monitoring performance, comprising a non-transitory computer storage medium in which computer program instructions are stored, which instructions, when executed by a processor, cause the processor to monitor application traffic comprising requests and responses of a single computer application that is carried in inbound and outbound packet traffic via at least one socket, the at least one socket excluding packet traffic of other computer applications, the inbound and outbound packet traffic being communicated over a data network, to monitor receipt times and exit times of data packets and respective packet headers to identify corresponding request packets and response packets of the single computer application among packets of the inbound and outbound packet traffic transiting the at least one socket, to define a performance indicator of the application traffic based on the receipt times and exit times of the corresponding request packets and response packets, to measure a first instance of the performance indicator during a first monitoring interval, to measure a second instance of the performance indicator during a second monitoring interval, to make a determination that a change between the first instance and the second instance exceeds a predetermined threshold, and responsively to the determination, to issue an alert.

14. The computer software product according to claim 13 , wherein the single computer application runs on a virtual machine.

15. The computer software product according to claim 13 , wherein the instructions cause the processor to measure the first instance and to measure the second instance only when the first instance exceeds a predetermined level.

16. The computer software product according to claim 13 , wherein the instructions cause the processor to adjust a computer resource responsively to the determination.

17. A data processing system for monitoring performance comprising: a memory for storing programs and data objects; and a processor, which is coupled to access the memory and is operative to perform the steps of: monitoring application traffic comprising requests and responses of a single computer application that is carried in inbound and outbound packet traffic transiting at least one socket over a data network, the at least one socket excluding packet traffic of other computer applications; monitoring receipt times and exit times of data packets and respective packet headers to identify corresponding request packets and response packets of the single computer application among packets of the inbound and outbound packet traffic transiting the at least one socket; defining a performance indicator of the application traffic based on the receipt times and exit times of the corresponding request packets and response packets; measuring a first instance of the performance indicator during a first monitoring interval; measuring a second instance of the performance indicator during a second monitoring interval; making a determination that a change between the first instance and the second instance exceeds a predetermined threshold; and responsively to the determination, issuing an alert.